A system for performing single Gaussian skin detection is described herein. The system includes a memory and a processor. The memory is configured to receive image data. The processor is coupled to the memory. The processor is to generate a single Gaussian skin model based on a skin dominant region associated with the image data and a single Gaussian non-skin model based on a second region associated with the image data and to classify individual pixels associated with the image data via a discriminative skin likelihood function based on the single Gaussian skin model and the single Gaussian non-skin model to generate skin label data associated with the image data.
Legal claims defining the scope of protection, as filed with the USPTO.
1. A system for performing single Gaussian skin detection, comprising: a memory configured to receive image data; and a processor coupled to the memory, the processor to: generate a single Gaussian skin model based on a skin dominant region associated with the image data and a single Gaussian non-skin model based on a second region associated with the image data; and classify individual pixels associated with the image data via a discriminative skin likelihood function based on the single Gaussian skin model and the single Gaussian non-skin model to generate skin label data associated with the image data.
2. The system of claim 1 , wherein classifying individual pixels comprises comparing a skin classification value for a first individual pixel to a predetermined threshold value.
3. The system of claim 1 , wherein the discriminative skin likelihood function comprises at least one of a log-based classifier, a thresholding function, a Bayesian classifier, a Gaussian classifier, a multi-layer perceptron classifier, or a neural network classifier.
4. The system of claim 1 , wherein the image data comprises an image frame of a video sequence, and wherein the processor to generate the skin model and the non-skin model and to classify the individual pixels comprises the processor to generate the skin model and the non-skin model and to classify the individual pixels online with respect to the video sequence.
5. The system of claim 1 , wherein the image data comprises an image frame of a video sequence and the processor is further to receive second image data associated with a second image frame of the video sequence, to generate a second skin model and a second non-skin model based on the second image frame, and to classify second individual pixels associated with the second image data via a second discriminative skin likelihood function based on the second skin model and the second non-skin model to generate second skin label data associated with the second image data.
6. The system of claim 1 , wherein the skin label data further comprises, for each of the individual pixels, a classification confidence value.
7. The system of claim 1 , wherein the processor is further to determine a minimum bounding box based on a plurality of facial feature landmarks and to expand the minimum bounding box to generate the skin dominant region.
8. The system of claim 1 , wherein the image data is in a color space comprising at least one of a red green blue color space, a luminance chroma color space, a luminance blue difference red difference color space, or a hue saturation value color space.
9. The system of claim 1 , wherein the processor is further to generate a second skin model based on a second skin dominant region associated with the image data and wherein the processor to classify the individual pixels associated with the image data further comprises the processor to classify the individual pixels as first skin pixels associated with the skin dominant region or second skin pixels associated with the second skin dominant region.
10. A method for performing skin detection, comprising: generating a single Gaussian skin model based on a skin dominant region of an image and a single Gaussian non-skin model based on a second region of the image; determining a discriminative skin likelihood function based on the a single Gaussian skin model and the single Gaussian non-skin model; and classifying individual pixels of the image as skin pixels or non-skin pixels based on the discriminative skin likelihood function to generate skin label data associated with the image.
11. The method of claim 10 , wherein classifying individual pixels comprises comparing a skin classification value for a first individual pixel to a predetermined threshold value.
12. The method of claim 10 , wherein the discriminative skin likelihood function comprises at least one of a log-based classifier, a thresholding function, a Bayesian classifier, a Gaussian classifier, a multi-layer perceptron classifier, or a neural network classifier.
13. The method of claim 10 , wherein the image comprises an image frame of a video sequence, and wherein generating the skin model and the non-skin model, determining the discriminative skin likelihood function, and classifying the individual pixels are performed online during processing of the video sequence.
14. The method of claim 10 , wherein the image comprises an image frame of a video sequence, the method further comprising: receiving a second image frame of the video sequence; generating a second skin model and a second non-skin model based on the second image frame; determining a second discriminative skin likelihood function based on the second skin model and the second non-skin model; and classifying second individual pixels of the second image frame based on the second discriminative skin likelihood function to generate second skin label data associated with the second image frame.
15. At least one non-transitory machine readable medium comprising a plurality of instructions that, in response to being executed on a computing device, cause the computing device to perform skin detection by: generating a single Gaussian skin model based on a skin dominant region of an image and a single Gaussian non-skin model based on a second region of the image; determining a discriminative skin likelihood function based on the single Gaussian skin model and the single Gaussian non-skin model; and classifying individual pixels of the image as skin pixels or non-skin pixels based on the discriminative skin likelihood function to generate skin label data associated with the image.
16. The non-transitory machine readable medium of claim 15 , wherein classifying individual pixels comprises comparing a skin classification value for a first individual pixel to a predetermined threshold value.
17. The non-transitory machine readable medium of claim 15 , wherein the discriminative skin likelihood function comprises at least one of a log-based classifier, a thresholding function, a Bayesian classifier, a Gaussian classifier, a multi-layer perceptron classifier, or a neural network classifier.
18. The non-transitory machine readable medium of claim 15 , wherein the image comprises an image frame of a video sequence, and wherein generating the skin model and the non-skin model, determining the discriminative skin likelihood function, and classifying the individual pixels are performed online during processing of the video sequence.
19. The non-transitory machine readable medium of claim 15 , further comprising instructions that, in response to being executed on the computing device, cause the computing device to perform skin detection by: determining a minimum bounding box based on a plurality of facial feature landmarks and expanding the minimum bounding box to generate the skin dominant region.
20. An apparatus for performing skin detection comprising: a controller for generating a single Gaussian skin model based on a skin dominant region of an image and a single Gaussian non-skin model based on a second region of the image; a likelihood unit for determining a discriminative skin likelihood function based on the single Gaussian skin model and the single Gaussian non-skin model; and a classifier unit for classifying individual pixels of the image as skin pixels or non-skin pixels based on the discriminative skin likelihood function to generate skin label data associated with the image.
21. The apparatus of claim 20 , wherein the classifier unit compares a skin classification value for a first individual pixel to a predetermined threshold value.
22. The apparatus of claim 20 , wherein the discriminative skin likelihood function comprises at least one of a log-based classifier, a thresholding function, a Bayesian classifier, a Gaussian classifier, a multi-layer perceptron classifier, or a neural network classifier.
23. The apparatus of claim 20 , wherein the image comprises an image frame of a video sequence, and wherein the means for generating the skin model and the non-skin model, the means for determining the discriminative skin likelihood function, and the means for classifying the individual pixels are to operate online with respect to the video sequence.
24. The apparatus of claim 20 , wherein the image comprises an image frame of a video sequence, the apparatus further comprising: a receiver for receiving a second image frame of the video sequence; generating a second skin model and a second non-skin model via the controller based on the second image frame; determining a second discriminative skin likelihood function via the likelihood unit based on the second skin model and the second non-skin model; and classifying second individual pixels of the second image frame via the classifier unit based on the second discriminative skin likelihood function to generate second skin label data associated with the second image frame.
25. The apparatus of claim 20 , wherein the skin label data further comprises, for each of the individual pixels, a classification confidence value.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
March 25, 2016
November 24, 2020
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.